IIUC, the NOP/verify pattern is primarily being justified due to bare usage. I am suggesting an alternative (at handwave height) which fits the same hole, while converting the rest of the functionality to something that is more efficient for the use cases I care about, signing hashes that reside on the stack.
I admit this is speaking off the cuff, but I’d like to know why this is not a good idea.
If consensus arrives that the bare case is worthwhile, just bundle them.
If reviewers (me) don’t have to think about legacy script, it’s a win.